ARIA: A Diagnostic Framework for Music Training Data Attribution
For researchers and practitioners in music AI and copyright analysis, ARIA provides a principled way to evaluate and interpret TDA methods, addressing the need for per-aspect attribution evidence aligned with the idea-expression distinction.
ARIA is a diagnostic framework for training data attribution (TDA) in music generation that decomposes influence along musical aspects (five for symbolic, three for audio) and provides reliability diagnostics. On a symbolic-music model with ground truth, it ranks four attribution methods identically to counterfactual retraining; on an audio model, it reveals varying attribution behaviors and flags issues like near-identical retrieval across queries.
Training data attribution (TDA) for music generation must answer two questions that copyright analysis requires, namely which training songs influence a generated output and along which musical aspects the influence operates. Existing methods reduce influence to a single scalar, without revealing which musical aspects are dominant in that influence. We propose ARIA, a framework that decomposes attribution along musical aspects (five for symbolic music, three for audio) and pairs the decomposition with reliability diagnostics computed from the segment-level score matrix. It measures within-group similarity among the top-K attributed tracks against random reference groups drawn from the training pool, and diagnoses the score matrix through its singular value decomposition and column statistics. On a symbolic-music model where attribution ground truth is available through counterfactual retraining, the reliability diagnostics rank four attribution methods identically to that ground truth. On an audio music generation model, ARIA reveals attribution behaviors that vary substantially across TDA methods, flags score matrices whose retrieved tracks are nearly identical across queries rather than reflecting per-query attribution, and characterizes embedding-similarity retrieval baselines by the musical aspect each encoder surfaces. Together, ARIA produces per-aspect attribution evidence aligned with the musical aspects considered under the idea-expression distinction in copyright analysis.